{"id":1849,"date":"2026-09-01T09:00:00","date_gmt":"2026-09-01T07:00:00","guid":{"rendered":"https:\/\/pletzenauer.com\/?p=1849"},"modified":"2026-07-27T17:48:54","modified_gmt":"2026-07-27T15:48:54","slug":"what-ai-really-costs","status":"publish","type":"post","link":"https:\/\/pletzenauer.com\/en\/2026\/09\/01\/what-ai-really-costs\/","title":{"rendered":"What AI really costs: 10 employees, four vendors, real numbers"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\"><strong>The short answer:<\/strong> for ten people the licence costs between roughly \u20ac1,600 and \u20ac2,500 a year \u2014 and it is the smallest of the three cost blocks. Rollout and training typically exceed it. Anyone comparing AI tools by seat price is optimising the smallest lever available, then wondering why the project got expensive anyway.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><em>Last checked: September 2026. Licence prices were verified against the vendors&#8217; own pages on the day of publication. The figures for rollout and training are my own experience values from consulting projects, not market data \u2014 they may differ in your organisation.<\/em><\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Disclosure:<\/strong> I earn my living from consulting and training around Claude \u2014 that is, from precisely the items that appear in this calculation as &#8220;rollout&#8221; and &#8220;training&#8221;. So read these numbers with particular scepticism. I have set them at the level I would defend in a quote.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">The short version<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Licence cost for 10 people:<\/strong> roughly \u20ac1,600 (Google Workspace Standard) to \u20ac2,520 (ChatGPT Business) per year.<\/li>\n<li><strong>Bundles contain more:<\/strong> with Google and Microsoft the office software is in the price. With Claude and ChatGPT it is not.<\/li>\n<li><strong>Rollout:<\/strong> a one-off \u20ac2,000 to \u20ac6,000 \u2014 depending on how tidy your document store is.<\/li>\n<li><strong>Training and support:<\/strong> \u20ac1,500 to \u20ac4,000 in the first year. The item most often cut, and the one that decides success or failure.<\/li>\n<li><strong>Realistic total in year one:<\/strong> two to four times the licence sum.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">The scope of the calculation<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">I am modelling a business I see often: <strong>ten people, all to get access, no in-house IT staff, an organically grown file store, no prior experience with AI tools in the business.<\/strong> Period: twelve months from the decision.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">One note on comparability that carries the whole calculation: with Microsoft and Google you are not buying an AI tool but office software <em>with<\/em> AI. You would have had that cost anyway. With Claude and ChatGPT you are buying an additional tool alongside your existing office software. Putting both in the same column compares apples with fruit baskets \u2014 so I show them separately.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Block 1: licence cost<\/h2>\n\n\n\n<figure class=\"wp-block-table\"><table><thead><tr><th>Option<\/th><th>per person\/month<\/th><th>10 people, 12 months<\/th><th>office software included?<\/th><\/tr><\/thead><tbody>\n<tr><td>Google Workspace Standard<\/td><td>\u20ac13.60<\/td><td><strong>\u20ac1,632<\/strong><\/td><td>yes<\/td><\/tr>\n<tr><td>M365 Business Standard with Copilot<\/td><td>\u20ac20.36<\/td><td><strong>\u20ac2,443<\/strong><\/td><td>yes<\/td><\/tr>\n<tr><td>M365 Copilot as add-on<\/td><td>\u20ac18.20 (standard)<\/td><td><strong>\u20ac2,184<\/strong> + base licence<\/td><td>no<\/td><\/tr>\n<tr><td>Claude Team<\/td><td>$20<\/td><td><strong>$2,400<\/strong> + office software<\/td><td>no<\/td><\/tr>\n<tr><td>ChatGPT Business<\/td><td>\u20ac21<\/td><td><strong>\u20ac2,520<\/strong> + office software<\/td><td>no<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">All figures on annual billing; monthly billing costs 20 to 25 per cent more depending on vendor. Anthropic publishes in US dollars, the others in euros \u2014 I deliberately do not convert. Microsoft&#8217;s promotional price of \u20ac15.60 applies only to contracts signed by 30 September 2026 and would bring that line down to \u20ac1,872.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>What to take from this table:<\/strong> the spread is less than \u20ac900 a year. Across ten people and twelve months that is around \u20ac7.50 per person per month \u2014 less than an hour of working time. A decision resting primarily on that difference is very probably the wrong one.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Block 2: rollout (one-off)<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">Here begins the part that appears in no quote comparison. What actually arises:<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Selection and testing<\/strong> \u2014 two to three weeks alongside normal work if you do it properly. Often not priced as internal effort, but real.<\/li>\n<li><strong>Setup and user management<\/strong> \u2014 manageable, usually half a day to a day.<\/li>\n<li><strong>Policy and data protection review<\/strong> \u2014 one to two days, more if a works council or external review is involved.<\/li>\n<li><strong>Cleaning up the document store<\/strong> \u2014 the great unknown. With Copilot practically unavoidable, because the assistant reaches into your existing store and in doing so exposes everything filed or shared wrongly.<\/li>\n<li><strong>Integrations with line-of-business systems<\/strong> \u2014 only if you need them, but then quickly several days.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>My experience value:<\/strong> \u20ac2,000 to \u20ac6,000 one-off. The lower bound applies to a business with a maintained document store introducing a standalone tool. The upper bound applies to a Copilot rollout in a SharePoint environment grown over years with historical permissions. I have seen both, and the difference never lay in the tool but in the starting position.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Block 3: training and support<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The item most often cut \u2014 and the only one that decides success or failure. A tool nobody uses is one hundred per cent wasted, however cheap the licence was.<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Basic training:<\/strong> two to three guided hours per person. Not as a lecture to everyone, but in small groups with real tasks from their own working day.<\/li>\n<li><strong>Support through the first three months:<\/strong> the period when the real questions surface. Without a named contact, the rollout peters out here.<\/li>\n<li><strong>Follow-up training after six months:<\/strong> optional but effective \u2014 by then people are far enough along that advanced topics actually land.<\/li>\n<\/ul>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>My experience value:<\/strong> \u20ac1,500 to \u20ac4,000 in the first year for ten people. The lower bound means basic training plus loose support; the upper a structured rollout over several weeks, as described in my <a href=\"https:\/\/pletzenauer.com\/en\/2026\/06\/11\/claude-desktop-team-rollout\/\">six-week plan<\/a>.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">The total<\/h2>\n\n\n\n<figure class=\"wp-block-table\"><table><thead><tr><th>Block<\/th><th>lean<\/th><th>solid<\/th><\/tr><\/thead><tbody>\n<tr><td>Licence (10 people, 12 months)<\/td><td>\u20ac1,632<\/td><td>\u20ac2,520<\/td><\/tr>\n<tr><td>Rollout (one-off)<\/td><td>\u20ac2,000<\/td><td>\u20ac6,000<\/td><\/tr>\n<tr><td>Training and support<\/td><td>\u20ac1,500<\/td><td>\u20ac4,000<\/td><\/tr>\n<tr><td><strong>Year one total<\/strong><\/td><td><strong>\u20ac5,132<\/strong><\/td><td><strong>\u20ac12,520<\/strong><\/td><\/tr>\n<tr><td>Licence share<\/td><td>32%<\/td><td>20%<\/td><\/tr>\n<tr><td>Subsequent years (licence + some support)<\/td><td>\u2248 \u20ac2,100<\/td><td>\u2248 \u20ac3,500<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">The decisive row is &#8220;licence share&#8221;: <strong>between 20 and 32 per cent.<\/strong> Four fifths of your first-year budget is therefore decided by questions that appear on no pricing page.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">And the second important row is the last one: from year two, rollout costs fall away. Viewed over three years the investment looks considerably friendlier \u2014 provided the rollout succeeded. Which is exactly why economising on the training line is the most expensive decision in this table.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">What this calculation does not include<\/h2>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Your own working time<\/strong> for selection, coordination and decision-making. In small businesses often the largest uncounted item.<\/li>\n<li><strong>Legal review<\/strong>, if you buy it in externally.<\/li>\n<li><strong>The productivity dip<\/strong> in the first weeks \u2014 people learning a new tool are temporarily slower.<\/li>\n<li><strong>Additional features<\/strong> needed only later: Google&#8217;s paid add-on, or with Claude Enterprise the usage billed at API rates above the seat price.<\/li>\n<li><strong>Exit costs<\/strong>, should you switch in two years.<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">When it pays off<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The honest answer: this cannot be calculated seriously in advance, and I distrust every table claiming otherwise. What I recommend instead is measuring one single, clearly defined process.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Example: writing a proposal currently takes you 90 minutes on average. After three months of use, measure again. If it is then 60 minutes and you write 20 proposals a month, you save ten hours monthly. Whether that pays off you can answer with your own hourly rate \u2014 using a number you measured rather than guessed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Crucially: measure <em>beforehand<\/em>. In hindsight, everyone remembers that it used to take longer.<\/p>\n\n\n\n<h2 class=\"wp-block-heading\">Frequently asked questions<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Can we save on rollout by doing it ourselves?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Yes, and in many businesses that is the right route \u2014 provided one person is explicitly allocated time for it. &#8220;Someone will fit it in&#8221; is not an allocation. The item does not disappear from the calculation; it simply moves from the invoice into your payroll.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Is it enough to equip half the workforce?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">As a start, yes, and it is often the wiser option: equip the people whose work clearly suits it first, then adjust after three months based on actual usage. Just avoid one thing \u2014 letting the other half fall back on personal accounts. Then you save on licences and buy yourself a data protection problem.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Why are your training figures so high?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Because two to three guided hours per person across ten people is already 20 to 30 hours, plus preparation and support. If the range seems high, check it against your own internal hourly rate \u2014 the number stays in the same order of magnitude. What I consider wrong is budgeting zero.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Is annual prepayment worth it?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Arithmetically yes, it saves 20 to 25 per cent. In the first year I would still pay monthly while you do not yet know whether the tool lands with your people. The premium is the price of the option to stop after three months without a loss.<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">What if we already have Workspace or Microsoft 365?<\/h3>\n\n\n\n<p class=\"wp-block-paragraph\">Then your licence block drops considerably \u2014 with Google potentially to zero, since basic Gemini features are included. The rollout and training blocks, however, stay exactly the same. Which is precisely why &#8220;it is already included&#8221; is not yet a rollout.<\/p>\n\n","protected":false},"excerpt":{"rendered":"<p>The licence accounts for only 20 to 32 per cent of the cost. A full twelve-month calculation \u2014 including the items missing from every quote.<\/p>\n","protected":false},"author":1,"featured_media":1992,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[17],"tags":[],"class_list":["post-1849","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-automatisierung"],"_links":{"self":[{"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/posts\/1849","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/comments?post=1849"}],"version-history":[{"count":1,"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/posts\/1849\/revisions"}],"predecessor-version":[{"id":1958,"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/posts\/1849\/revisions\/1958"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/media\/1992"}],"wp:attachment":[{"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/media?parent=1849"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/categories?post=1849"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/pletzenauer.com\/en\/wp-json\/wp\/v2\/tags?post=1849"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}